Undergraduates short-form video addiction and learning burnout association involving anxiety symptoms and coping styles moderation

The rapid rise of short-form video platforms (e.g., TikTok/Douyin) has led to widespread use among Chinese university students, facilitated by recommendation algorithms and micro-length reward structures. These affordances can drive addiction-like use and resource depletion per Conservation of Resources (COR) theory, potentially impairing academic functioning and contributing to learning burnout. Drawing on COR and Job Demands-Resources (JD-R) frameworks, the study posits that excessive short-form video use depletes temporal and cognitive resources, elevates anxiety, and fosters burnout. Coping styles, conceptualized by the Stress Coping Model (SCM), may moderate these relationships. The study tests: H1, short-form video addiction positively relates to learning burnout; H2, anxiety symptoms mediate the addiction–burnout relationship; H3, positive coping styles moderate the direct path from addiction to burnout; H4, negative coping styles moderate the direct path from addiction to burnout.

Prior work links generalized internet addiction to burnout and poorer academic outcomes, with meta-analytic evidence of positive associations between problematic internet use and burnout. Short-form video platforms uniquely reinforce engagement through algorithmic personalization and variable reward schedules, contributing to attentional fragmentation and cognitive load. Research shows associations between short-form video addiction and stress, anxiety, reduced engagement, procrastination, and perceived learning ineffectiveness. COR theory frames addiction as resource loss (time, cognition, emotion), while JD-R suggests that mismatched demands and resources drive burnout. SCM highlights the role of coping strategies—positive coping (problem-solving, seeking support) may buffer stress, while negative coping (avoidance/withdrawal) may exacerbate maladaptive outcomes. Despite initial evidence, mechanisms linking short-form video addiction to learning burnout in undergraduates remain underexplored, motivating the current moderated mediation investigation.

Design: Cross-sectional survey employing validated scales and regression-based mediation and moderated mediation analyses. Participants and procedure: Convenience sampling of freshmen to seniors from a college in Jiangsu Province, China; online questionnaires yielded 523 valid responses (validity rate 97.76%). Demographics: 61.76% female (n=323), 38.24% male (n=200); 254 freshmen, 109 sophomores, 115 juniors, 45 seniors; 345 science/engineering, 116 humanities/social sciences, 62 other. Ethics: Approved by the Ethics Committee of Changshu Institute of Technology (KY0015); anonymity and informed consent ensured. Measures: (1) Coping styles—Simple Coping Style Scale (Xie, 20 items; positive coping items 1–12, negative coping items 13–20; 4-point scale: 1 not adopted to 4 often adopted). Reliability: total α=0.921; positive coping α=0.932; negative coping α=0.824. (2) Short-form video addiction—8-item scale adapted from Zhang et al.; 7-point Likert (1 very disagree to 7 very agree); α=0.923. (3) Learning burnout—Lian et al. 20-item college student scale, three dimensions (low mood, improper behavior, low sense of accomplishment); 5-point Likert; reverse-scored items 1,3,6,8,11,13,15,18; total α=0.883; dimension αs=0.881, 0.734, 0.746. (4) Anxiety symptoms—GAD-7; 7 items; 0–3 Likert; α=0.933. Data analysis: Psychometric adequacy checked (Nunnally & Bernstein criteria; Harris 20:1 ratio; Aiken & West interaction guidance). EFA with Principal Axis Factoring and Direct Oblimin rotation showed loadings 0.351–0.745, meeting retention thresholds. Harman single-factor test indicated 20.605% variance, suggesting limited common method bias. Descriptive statistics and Pearson correlations computed. All continuous variables standardized. Mediation tested via regression; moderated mediation via PROCESS v3.5 (Model 5). Bias-corrected bootstrap (5000 resamples) assessed indirect effects (90% CI).

Descriptive statistics (N=523): Means (SD)—SFVA 3.027 (1.303), LB 2.944 (0.374), AS 0.895 (0.686), PCS 2.804 (0.583), NCS 2.325 (0.560). Correlations: SFVA–LB r=0.280 (p<0.01); SFVA–AS r=0.447 (p<0.01); AS–LB r=0.201 (p<0.01); PCS negatively correlated with SFVA (r=-0.220, p<0.01) and AS (r=-0.197, p<0.01), and weakly with LB (r=-0.181, ns); NCS positively correlated with SFVA (r=0.216, p<0.01), AS (r=0.304, p<0.01), LB (r=0.159, p<0.01). Mediation: Equation 1 (LB ~ SFVA): β=0.280, t=6.65, p<0.001, 95%CI [0.057, 0.104], R²=0.078, F=44.226***. Equation 2 (AS ~ SFVA): β=0.447, t=11.391, p<0.001, 95%CI [0.195, 0.276], R²=0.199, F=129.762***. Equation 3 (LB ~ SFVA + AS): SFVA β=0.237, t=5.061, p<0.001, 95%CI [0.042, 0.095]; AS β=0.095, t=2.033, p=0.043, 95%CI [0.002, 0.102]; R²=0.086, F=24.312***. Indirect effect via AS: 0.043 (0.447 × 0.095), accounting for 15.4% of total effect; bootstrap 90% CI [0.004, 0.083] significant. Moderated mediation (PROCESS Model 5): LB predicted by SFVA β=0.089, t=5.041, p<0.001, 95%CI [0.054, 0.124]; AS β=0.030, t=1.652, p=0.099; NCS β=0.030, t=1.822, p=0.069; interaction SFVA×NCS β=-0.040, t=-2.813, p=0.005, 95%CI [-0.068, -0.012]; model R²=0.106, F=15.367***. Simple slopes by NCS level: Low NCS slope=0.130 (SE=0.023), t=5.540, p<0.001, 95%CI [0.084, 0.176]; Mean NCS slope=0.089 (SE=0.018), t=5.041, p<0.001, 95%CI [0.054, 0.124]; High NCS slope=0.049 (SE=0.022), t=2.193, p=0.029, 95%CI [0.005, 0.092]. Positive coping styles did not significantly moderate the direct path.

Findings support H1 and H2: short-form video addiction positively relates to learning burnout and elevates anxiety symptoms, with anxiety partially mediating the addiction–burnout link. These results align with COR and JD-R models: compulsive use depletes temporal and cognitive resources, increases anxiety, and contributes to burnout via resource–demand imbalance. Contrary to expectations, H4 showed that higher negative coping styles weakened the direct association between addiction and burnout (interaction β=-0.040), suggesting that avoidance-type coping may dampen the immediate impact of addiction on burnout, potentially by temporarily diverting attention from academic stressors. H3 was not supported: positive coping styles did not moderate the direct path, possibly due to self-control resource depletion limiting the capacity to deploy positive strategies under compulsive media use. The integrated COR→JD-R→SCM framework offers a mechanistic account of how short-form video addiction impacts academic outcomes, emphasizing anxiety as a pathway and coping styles as boundary conditions.

This study identifies a dual-path mechanism linking short-form video addiction to learning burnout among Chinese undergraduates: a direct positive association and an indirect pathway via increased anxiety symptoms (indirect effect=0.043, 15.4% of total effect). Negative coping styles uniquely buffered the direct addiction–burnout link. Practical recommendations include embedding algorithm-awareness content in digital literacy curricula, strengthening campus mental health services and peer support networks, and conducting scenario-based workshops to cultivate adaptive coping skills. Future research should employ longitudinal or experimental designs, incorporate objective usage metrics, and examine additional psychological resources (e.g., resilience, self-efficacy) to refine intervention targets.

Self-report measures may introduce common method bias; although Harman’s test suggested limited bias, future work should use multi-source or pairwise designs and objective usage metrics. Cross-sectional data preclude causal inference and exploration of bidirectionality; learning burnout may also predict short-form video addiction. Convenience sampling from a single institution limits generalizability; samples beyond college students, including adolescents, are needed. The study focused on anxiety and coping styles; other psychological resources (e.g., resilience, self-efficacy) likely influence addiction and burnout and warrant inclusion in extended models.